INTRODUCTION TO MACHINE LEARNING

$1,000,000

Machine Learning  Learn models from data  Three main types of learning :  Supervised learning  Unsupervised learning  Reinforcement learning

Variants of Machine Learning Problems  What is being learned ?  Parameters, problem structure, hidden concepts, …  What information do we learn from ?  Labeled data, unlabeled data, rewards  What is the goal of learning ?  Prediction, diagnostics, summarization, …  How do we learn ?  Passive / active, online / offline  Outputs  Binary, discrete, continuous

Supervised Learning  Given a set of data points with an outcome, create a model to describe them  Classification  outcome is a discrete variable ( typically <10 outcomes )  Linear regression  outcome is continuous

Training Data Training set. Data set. Training data. Observations. InputOutput

What else do we have ?  In real life, we usually don ’ t have just a set of data  Also have background knowledge, theory about the underlying processes, etc.  We will assume just the data ( this is called inductive learning )  Cleaner and a good base case  More complex mechanisms needed to reason with prior knowledge

Inductive Learning

How can we pick the right function ?  There could be multiple models to generate the data  Examples do not completely describe the function  Search space is large  Would we know if we got the right answer ? Not the right way of thinking about the problem.

Inductive Learning

Avoiding Overfitting the Model 1. Divide the data that you have into a distinct training set and test set. 2. Use only the training set to train your model. 3. Verify performance using the test set. Measure error rate  Drawback of this method : the data withheld for the test set is not used for training  split of data means we didn ’ t train on half the data  split means we might not get a good idea of the accuracy

K - fold Cross - Validation

More Data Is Usually Better

Classification Algorithms  Decision trees  Neural networks  Logistic regression  Naïve Bayes  …

Decision Trees  …similar to a game of 20 questions  Decision trees are powerful and popular tools for classification and prediction.  Decision trees represent rules, which can be understood by humans and used in knowledge system such as database.

Learning decision trees Problem : decide whether to wait for a table at a restaurant, based on the following attributes : 1. Alternate : is there an alternative restaurant nearby ? 2. Bar : is there a comfortable bar area to wait in ? 3. Fri / Sat : is today Friday or Saturday ? 4. Hungry : are we hungry ? 5. Patrons : number of people in the restaurant ( None, Some, Full ) 6. Price : price range ($, $$, $$$) 7. Raining : is it raining outside ? 8. Reservation : have we made a reservation ? 9. Type : kind of restaurant ( French, Italian, Thai, Burger ) 10. WaitEstimate : estimated waiting time (0-10, 10-30, , >60)

Attribute - based representations  Examples described by attribute values ( Boolean, discrete, continuous )  E. g., situations where I will / won ' t wait for a table :  Classification of examples is positive ( T ) or negative ( F )

Decision trees  One possible representation for hypotheses

Expressiveness  Decision trees can express any function of the input attributes.  E. g., for Boolean functions, truth table row → path to leaf :  Trivially, there is a consistent decision tree for any training set with one path to leaf for each example ( unless f nondeterministic in x ) but it probably won ' t generalize to new examples  Prefer to find more compact decision trees

Decision tree learning  Aim : find a small tree consistent with the training examples  Idea : ( recursively ) choose " most significant " attribute as root of ( sub ) tree

Decision trees  One possible representation for hypotheses

Choosing an attribute  Idea : a good attribute splits the examples into subsets that are ( ideally ) " all positive " or " all negative "  Which is a better choice ?  Patrons